Introduction

Gordon et al. 2017 present a 3D continuum radiative transfer benchmark problem composed of a slab of dust with uniform density externally illuminated by a star. The aim is to test dust absorption, scattering, and emission, optionally taking into account stochastic heating of dust grains in non-LTE conditions (see Stochastically heated dust grains). The authors provide results produced by seven different radiative transfer codes, including SKIRT, for several optical depths and viewing angles.

References and downloads

Publication	Gordon et al. 2017 [ADS]
Ski file	gordon_template.ski
Wavelength grid - basic	gordon_wavelengths_basic.dat
Wavelength grid - full	gordon_wavelengths_full.dat

SKIRT results

The diagram below on the left illustrates the benchmark setup including the viewing angles. The figure on the right shows the SEDs calculated taking into account stochastic heating for all optical depth values in the benchmark and for selected viewing angles.

The figure below shows images produced by SKIRT for the $τ_{z} (1 μ m) = 0.1$ case calculated taking into account stochastic heating for two representative wavelengths.

Many more results and comparisons between the participating codes, including SKIRT, are presented in the benchmark paper (see References and downloads). The figures below present a comparison of the results produced by a recent version of SKIRT with those of three other codes participating in the benchmark for a specific benchmark configuration (stochastic heating, optical depth $τ_{z} (1 μ m) = 0.1$ , viewing angle $θ = 180^{\circ}$ , and for the images, wavelength $λ = 0.15 μ m$ ).

Performing this benchmark

To perform this benchmark, download the ski file and wavelength data files provided above (References and downloads). Open the ski file in a text editor to adjust the following parameter values to a particular benchmark configuration:

Parameter	XML element	XML attribute
Emission type	`DustEmissionOptions`	dustEmissionType
Wavelength grid	`FileWavelengthGrid` (5 times)	filename
Optical depth	`OpticalDepthMaterialNormalization`	opticalDepth
Number of dust grid bins along Z-axis	`meshZ/` `LinMesh`	numBins
Number of pixels in the instrument frames	`FullInstrument` (7 times)	numPixelsX ; numPixelsY
Nr of photon packets	`MonteCarloSimulation`	numPackets

The emission type can be one of two values, with a corresponding value for the wavelength grid (the spectral features resulting from stochastic heating would not be resolved on the lower resolution "basic" wavelength grid):

Emission type	Wavelength grid	Description
Equilibrium	gordon_wavelengths_basic.dat	Assume dust grains to be in equilibrium with radiation field
Stochastic	gordon_wavelengths_full.dat	Take into account stochastic heating of smaller dust grains

To ensure correct results for the benchmark configuration with the highest optical depth, the dust grid resolution must be increased. At the same time, the instrument resolution is increased to properly show the resulting surface brightness gradients. The appropriate values are:

Optical depth	Dust grid bins along Z-axis	Pixels in instrument frames
0.01	31	300
0.1	31	300
1	31	300
10	201	600

Finally, the number of photon packets should be adjusted depending on the benchmark configuration. The table below lists the values used for producing "official" benchmark results. For the higher optical depths, these values lead to long execution times.

Optical depth	Equilibrium	Stochastic
0.01	1e10	2e10
0.1	1e10	2e10
1	3e10	6e10
10	10e10	20e10

Finally, pass the (name of) the ski file to SKIRT as a single command line argument. At the end of the simulation run, SKIRT outputs data cubes and SEDs which can be compared to the original benchmark results.